Integrated energy monitoring and visualization system for smart green
Upcoming SlideShare
Loading in...5

Integrated energy monitoring and visualization system for smart green






Total Views
Views on SlideShare
Embed Views



0 Embeds 0

No embeds



Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
Post Comment
Edit your comment

    Integrated energy monitoring and visualization system for smart green Integrated energy monitoring and visualization system for smart green Document Transcript

    • AUTCON-01309; No of Pages 9 Automation in Construction xxx (2011) xxx–xxx Contents lists available at ScienceDirect Automation in Construction j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / a u t c o nIntegrated energy monitoring and visualization system for Smart GreenCity developmentDesigning a spatial information integrated energy monitoring model in the contextof massive data management on a web based platformSung Ah Kim a, Dongyoun Shin b,⁎, Yoon Choe a, Thomas Seibert b, Steffen P. Walz ca Sunkyunkwan University, Republic of Koreab ETH Zurich, Switzerlandc Royal Melbourne Institute of Technology, Australiaa r t i c l e i n f o a b s t r a c tArticle history: U-Eco City is a research and development project initiated by the Korean government. The projectsAccepted 18 July 2011 objective is the monitoring and visualization of aggregated and real time states of various energy usagesAvailable online xxxx represented by location-based sensor data accrued from city to building scale. The platforms middleware will retrieve geospatial data from a GIS database and sensor data from the individual sensory installedKeywords: over the city and provide the browser-based client with the accommodated information suitable toEnergy monitoringData visualization display geo-location characteristics specific to the respective energy usage. The client will be capable ofSmart Green City processing and displaying real time and aggregated data in different dimensions such as time, location,Spatial information model level of detail, mode of visualization, etc. The platforms middleware has been developed into anEnerISS operative, advanced prototype, providing information to a Web-based client that integrates andSocial sensing interfaces with the Google Earth and Google Maps plug-ins for geospatially referenced energy usage visualization and monitoring. © 2011 Elsevier B.V. All rights reserved.1. Introduction (EnerISS Viewer in Fig. 1). It is a Web-based rich-internet application (RIA) integrating a 3D geospatial viewer based on the Google Earth The objective of this research is to build an energy management platform and Google Maps components with additional data visual-system called EnerISS (Energy Integrated Urban Planning & Managing ization modules. Combining city information model data from theSupport System). The research initiated from the idea that the city can database, energy consumption data (through a sensor network), andprovide a systematic infrastructure that delivers extra power such as environmental GIS data, it builds up its own monitoring database. Theelectricity or heat to demanding areas through an intelligent power EnerGIS tool is powered by its middleware engine that has beennetwork. In this context, building up an effective energy-monitoring conceptually designed and developed by our project partner, atool is one of the most important tasks to implement for the EnerISS software development and strategic consulting company specializingscenario. The basic assumption is that the system is not only used by in geospatial systems, location-based as well as serious game management specialists but also by the general energy In the long run, this system will also provide an intelligent analysisconsumer. We deliver the information both to the non-professional and prescription system based on a business rule engine so thatuser and professional energy manager to enhance the notion of significant decisions for energy management can be suggestedenergy-aware consuming and to support more strategic and efficient through user-friendly interfaces meeting multiple viewpoints fromenergy usage decisions. Furthermore, by accumulating periodical various stakeholders including, the system will also be able to provide a statistical foundation to Fig. 1 illustrates the system architecture of the EnerISS solutionpredict and to design the future energy plan. In this context, we which integrates modeler, viewer (EnerGIS), database, and datapresent a prototype of an energy monitoring system called EnerGIS acquisition module. The EMS (Energy Management System) will control the operation of power plants based on the rules generated from the interaction between the solver and evaluator. They can ⁎ Corresponding author. Tel.: + 41 079 799 8476; fax: +41 44 633 15 75. predict the energy demand from E-GIS data, enabling in-time E-mail address: (D. Shin). planning of energy supply through the EMS.0926-5805/$ – see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.autcon.2011.07.004 Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004
    • 2 S.A. Kim et al. / Automation in Construction xxx (2011) xxx–xxx ambient lighting [5]. IBM has announced recently that it will soon publish a SimCity-like serious game named CityOne [11] that will allow to process real-world data and to “play out” possible and future results — an approach we assume will demonstrate how games as motivational and scenario systems can be taken advantage of to tackle real-world challenges and to let people participate in solving these challenges. In comparison, our client and middleware prototypes already integrate typical social game mechanisms such as leader-, and loserboard functionalities, to let users find out how their geospatial region has used energy over time. More game-like features are planned for future iterations. 3. Research objectives and method It is crucial to involve citizens who consume energy in order toFig. 1. EnerISS Solution, which has the following four different modules, interacts with build up an energy-ecosystem consisting of energy suppliers, energythe urban spatial information database, E-GIS, SCADA database and the energy source managers, policy makers and citizens [7]. Constructing this partici-unit/valuation information database: Modeler, Solver, Evaluator and EMS. patory environment, we can eventually convince energy consumers to achieve the targeted level of energy saving and reduced CO2 emission. With these demanding specifications, we identified the following The modeler allows interactive 3D modeling of urban space, and characteristics for our solution:automatically transfers the model data to the solver, and the solverestimates energy demand of the city by combining the e-GIS data and • Web based platformthe geospatial data. The evaluator, which analyzes energy usage to • Intuitive statistical data visualizationdeliver the most effective energy supply strategies, suggests feasible • Real-time based sensor data collection and data aggregationenergy plans by interacting with the solver/evaluator/EMS integrated • Dynamic data loading and visualizationdatabase. Meanwhile, the viewer visualizes all these information in • Extensible city informationreal time (Fig. 1). We used the 3D urban environment in the Google Earth plugin and designed various visualization alternatives to support various user2. Related work scenarios. In this context, we found critical research issues with regard to the system we wanted to achieve, defining the following research The most relevant work in IT technologies is the “smart grid [9]” objectives:that includes an intelligent monitoring system keeping track of allelectricity flowing in the system [1,2,4]. The technologies enhancing • Optimized data structure model of urban environmentenergy saving can be divided into the following two areas. The first • Data optimization for 3D city representationarea only informs about energy consumption data related with usage • Visualization strategypatterns in order to motivate citizens to save energy (smart meter,Google PowerMeter [10]). The second area does not only monitor the 3.1. Urban data structure modelenergy consumption but also controls the power source in a veryinteractive way [3,6]. A further area is, nowadays, expanding further The existing GIS system and diverse Building Information Modelto cover the life and public design in very emotional aspects like (BIM) technologies can represent the 3D geological environment with Fig. 2. EnerISS modeler supports diverse editing functions optimizing the urban spatial information models. Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004
    • S.A. Kim et al. / Automation in Construction xxx (2011) xxx–xxx 3Fig. 3. Several concept types for cell data visualization. The images show different types of cell data representation with the following attributes: color, height, 3D-geometry andalpha value. Also note that the Google Earth visualizations are always accompanied by chart representations that allow for representations of energy usage over time.semantic information, and we might utilize those existing technologies have detailed information on buildings are also not suitable to be easilyfor a small test bed environment. However, there are distinctive adopted because of the closed nature of their data querying character-limitations to the adoption of the technologies into our energy istics, since the data would have to be retrieved on a selective basis duemonitoring system, because the GIS system uses a pre-made 3D to the huge amount of information at the city scale. So we had to designmodel that cannot be easily extended and modified after data is an alternative solution, which is sustainable for ubiquitous (WWW)imported into our system. Additionally other BIM technologies that access of future users.Fig. 4. LOD scheme by KML 1. Building model in grid cell, grid cell has link to each building. 2. Building unit: base area × number of floors, each building has link to floors. 3. Floor levelof detail. Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004
    • 4 S.A. Kim et al. / Automation in Construction xxx (2011) xxx–xxx3.2. Data optimization for 3D city representation information and mash-up services, which contain diverse multimedia information [8]. In this context this system is designed to support a This simulation platform deals with massive data volumes due to dynamic 3D city environment, which is based on the Google Earththe huge number of buildings, the city environment model and the plugin. Delivering energy usage data in a complex city modelongoing data collection within short periods of time. In order to deliver especially requires well-organized spatial structure in accordancelarge amounts of GIS data and build the model more efficiently, we with different LODs, because using various view-dependent LODsadopted the Google Earth plugin, which is a virtual globe, map and helps to decrease the complexity of the 3D urban environmentgeographic information program that was originally called Earth- representation as it moves away from the viewer or according to otherViewer 3D created by Keyhole, Inc. The main data format processed by metrics such as area, block and building importance, eye-space speedGoogle Earth is named KML (Keyhole markup language). It has been or position. Therefore, we planned to set up five different levels ofsubmitted as a standard proposal to the Open Geospatial Consortium detail: area, block, building, floor, and unit, and for this test bed(OGC) and – since version 2.2 – has become an official standard. platform, we simplified these 5 LODs into 4 LODs that are grid, cell, The EnerISS Modeler allows parametric generation of 3D building building and floor (Fig. 4). “Block” level of detail originated from the e-models from GIS data with the option of manual editing. The KML file GIS system that is based on mesh cell and consists of 200 m × 200 m.was extended to accommodate additional attributes and LOD control This grid cell represents the wide-area of urban scale, like a pixel map.scripts when it was conceived at the beginning (Fig. 2). However, the When we navigate into the building scale, the “building” level of detailfinal design of the system maintained the original format without is activated automatically and shows each buildings energy usage byadditional information as the middleware engine generates scene different colors. Accordingly, zooming into the “floor” level brings upmodes in response to the user interaction. In this way, the visualized details on the floor scale. The “floor” level of detail that is the highestdata has a small footprint and does not require complicated interpre- LOD is interactively shown by user request when the mouse is doubletation of scripts on the client side. clicked on the aimed building in order to optimize real-time data flow which has relatively large amounts of data depending on the building3.3. Visualization strategy scale. Corresponding with these scenarios, we suggested “a middleware Citizen participation scenarios in energy saving is one of the main engine”, which aggregates the sensor data and city information tosubjects of this project from the initial moment. It requires an build up an optimized database that allows for fast and efficientintuitive, easy-to-use interface and a suitable representation method generating of dynamic visualization from web based requests (Fig. 5).to assist user interaction with the system. Therefore, diverse ways of The middleware engine has a database and a visualizationvisualization have been tried (Fig. 3) and several effective ways of component. All of the consumption and environmental data, whichrepresentation were implemented. will be provided by the sensor network, are accepted through a Restful interface (Representational State Transfer) and aggregated4. Implementation into the middleware engine, and the data processing component placed in the middleware engine retrieves the sensor data to4.1. Overview of design of energy visualization system aggregate them continuously. The server will combine geospatial data with energy consumption and environmental data in real time, The new trend for wide area geospatial information systems is web transform the raw data into the KML dialect and deliver it to the clientbased three-dimensional city models interacting with geospatial in a compressed format (KMZ files). Therefore when the end user Fig. 5. Middleware engine components and connections to other services. Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004
    • S.A. Kim et al. / Automation in Construction xxx (2011) xxx–xxx 5 Fig. 6. The pilot system of visualizing real time sensor data using the Google Earth plugin.requests the data, the KML service API, which is embedded in a web- and optimize the collected data in order to be able to deal with thebrowser, can directly deliver the data set from the server, without massive amount of data being fed into the system over time. This hascomplex data processing, and visualizes it immediately. to be done by designing aggregated table or data cubes and processing the data coming from upstream systems in an asynchronous way. The system therefore has to implement the typical functionalities of a data4.2. Test bed for sensor data visualization in the Google earth component warehouse. Before we practically started to develop Urban Integrated EnegySystem (UIES) platform, we made a pilot system to test several 4.3. Urban data structure model for energy visualizationtechnical issues that mainly concern real time sensor network andsome representation method with the Google Earth plugin (Figs. 6, 7). Implementation of the energy visualization via a web based clientTen wireless environment sensors were installed in a building that is which dynamically supports loading the preprocessed data and serveslocated in Dongtan Korea, and collected the following data: CO2 information visualization in real time by calling of complex objectivesdensity, temperature and humidity in real-time. From this pilot fully and consistently will realize our ambitious concept. A well-platform, we found several issues concerning wireless sensor designed data structure of the urban environment enhances thenetworks, extensive data aggregation and the visualization within system performance and data representation method (Fig. 8). Whenthe Google Earth plugin. we built up the system, it seemed not suitable to simply utilize First, the wireless network between sensors sometimes causes existing urban data models because of the data optimization issue.trouble inside buildings due to the walls blocking signals. Therefore it Therefore we built up the data structure with UIES customizedis crucial to analyze the internal building characteristics before attributes. In the geospatial dimension, we worked with severalinstalling sensors. Second, with regard to the periodical data that entities that match the various levels of details (LODs). Each entitywill be collected by the sensors, we needed to constantly aggregate apart from the highest one in the hierarchy has a parent entity (e.g. a Fig. 7. Raw sensing data is visualized in diverse ways: a bar graph and line graph, daily, hourly, and latest time selections. Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004
    • 6 S.A. Kim et al. / Automation in Construction xxx (2011) xxx–xxx Fig. 8. Optimized data structure for urban spatial information.cell entity consists of several building entities, which all have the cell respective sensor node, from which the data was coming and harborsentity as parent) in order to be capable of performance and precise properties such as the time at which the values has been captured, thedata traversal up and down the hierarchy. Every geospatial entity value itself and the type of the value. This design makes it possible todisposes of a Bounding Box object, so that it can be clearly identified have strong referential integrity for the accrued data and beby its most simple geospatial attributes (i.e. latitude and longitude for expandable for future requirements such as more kinds of dataall 4 sides of a proposed rectangle). coming from the sensor network. Despite their similar structure, we have decided to store consump-tion and environmental values in two different tables due to the 4.4. Large data treatmentnecessity of performance when dealing with huge amounts of data(within our tests we realized that especially the consumption values After our first stress tests, which were limited to data generation foraggregated to thousands of rows within just a few minutes). The just one grid (the test bed area as described above) we realized that themeasurement values represent the data provided from the sensor performance of querying data on a real-time basis below the grid levelnetwork. Therefore each sensor value has an association to the will very soon become unacceptable for the end user, even though the Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004
    • S.A. Kim et al. / Automation in Construction xxx (2011) xxx–xxx 7Fig. 9. Level of details: “mesh grid” is divided by e-GIS system unit that is based on mesh cell and consist of 200 by 200 m2. “Building” LOD defines each building unit and “Floor” LODdefines each floor.established mechanism of dynamically reloading data in real-time following four steps: mesh grid, block, building and floor levels. Theaccording to user actions helped in the beginning to limit data traffic to given LOD, mesh grid, block, and building, specify physical boundariesthe data which was actively requested by the users. With consumption in an urban area. The “block” level of details considers semanticsensor data possibly being provided every 10 min and environmental regional boundaries in order to analyze customized area (Fig. 9).data being updated every hour, the amount of data collected over the As the distance to the earths surface becomes closer, the clientcourse of a single day was so huge that the request of this data let alone automatically switches to a higher LOD (here the building or floorthe real-time preprocessing (algorithms building averages, ratios or level) and buildings are displayed three-dimensionally in a specificaccumulations) forced the browser application and their respective color, with the color red denoting a consumption relatively located inAJAX functionality to slow down in an unacceptable manner. the upper 33%, yellow for a consumption value located in the middle The time dimension also mandated a powerful mechanism on the 33% and green for a value in the lower 33% (Fig. 10).server side to aggregate the incoming data feed, especially when timegranularity was decreased (up to year level). Therefore the middleware 4.5.2. Display of charts for a selected data typehad to utilize a mining component, calling this component whenever When a data type has been selected and a cell on the map is doublenew data arrived, so that the mining component could process the clicked, the user can have the corresponding values displayed in timeincoming data and populate the aggregation tables accordingly. Data via a line diagram. The bottommost panel on the left side of thisprocessing of the mining component was done in an asynchronous way section lets the user put those values in correlation. The last 2 clickedso that performance of accepting and collecting sensor data from the items will be brought in correlation and be displayed in the third andsensor network would not be affected by the resource consuming task of bottommost line diagram. The following chart shows a simplebuilding data suited for the aggregation tables. The aggregation tables correlation between two selected factors (Fig. 11). In the diagrams awere build for 2 dimensions: the LOD-dimension, where data was user can juxtapose two distinct series of consumption and environ-aggregated on building and cell level, and a time-based dimension mental data, with the values for consumption data over time shown inwhere data was aggregated per hour, day and month. With this the topmost diagram and the values for environmental data over timewarehoused data the front-end application was able to request data shown in the middle diagram. A trend analysis is then performed,from a specific region, with the server side components balancing out which puts the selected environmental data curve into correlationthe best way to collect and deliver the requested information. with the selected consumption data curve. Thus the bottom diagram shows the correlation intensity between the two selected curves4.5. Representation strategy with diverse visualization above: the higher the value in the third diagram, the stronger the correlation between the two selected values above will be. In the4.5.1. Choropleth map in 3D urban environment example shown in Fig. 11, it is clearly displayed that the correlation According to the current distance to the earths surface all cell data between humidity and heat consumption gets lower as the twoare displayed as a choropleth map. We defined the LOD in the topmost curves head for different directions. The correlation analysis Fig. 10. Choropleth map in different levels of detail: block, building and floor. Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004
    • 8 S.A. Kim et al. / Automation in Construction xxx (2011) xxx–xxx Fig. 11. Line diagrams for the different data type series and their correlations.can be performed between each environmental and consumption diverse social networking services (Twitter, Facebook, etc.) into thedata series simply by clicking on the symbols to the left of the EnerISS solution that enables interaction with the public, and it will alsodiagram. be enhanced by a human centered sensing environment. Furthermore, by implementing a business rule engine, we will notonly execute data correlation between sensor and environment data 5. Conclusionbut also expect to extract meaningful mining output that we didntassume before. Currently, we design our own database structure to enhance system performance and to implement a customized middleware4.5.3. Diverse representations engine to optimize the visualization method, which is mainly based If users prefer to have the leaderboards shown in the course of on a LOD strategy. In this process, we acquired knowledge fortime (with monthly granularity), they can select a tabular view and databases that manage large amounts of data that are continuouslyget forwarded to a different web page that displays both the aggregated by time flow, and for a system architecture that representsleaderboard for best performance as well as the leaderboard for the an energy driven urban environment. Furthermore we acquired theworst performance over the last year for the selected data type. If the know-how for a web based management system, which deliversuser selects to have the leaderboard displayed in the Google Earth sensor data, statistical data mining and diverse informationplug-in, a pillar for each cell is displayed, with the height and the color visualization.both denoting the rank in the leaderboard. The tallest pillars arealways green colored and show the “leading” cells, whereas the 6. Future worklowest pillars are colored red and show the cells, which consumptionvalue has evolved the worst within the given month (Fig. 12). This research represents an energy monitoring and visualization system and will implement an interactive energy management4.5.4. Integration of social sensors system for the next step. Based on the current development, the So far, we mainly used mash-up methods to support better decision mining engine, which supports comparison between selected factors,making in order to manage an effective energy supply and demand level will be improved by implementing a business rule engine, which canwhich is based on analyzing temperature, humidity, time and other automatically discover significant knowledge from the database.crucial values. Meanwhile, Fig. 13 shows another trial which displays Furthermore, the LOD capabilities will be elaborated from floor levelsensing data by the mash up method. We are planning to implement to unit (room) level, which will empower the system to evaluate Fig. 12. Leader as tabular view and leaderboard view within the Google Earth plugin for Electricity. Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004
    • S.A. Kim et al. / Automation in Construction xxx (2011) xxx–xxx 9 Fig. 13. Mash-up visualization of the social sensing efficiency respect to the building direction, materials and other [2] M. Amin, National infrastructures as complex interactive networks. (2000) Chapter 14: 263–286.environmental factors. In addition, it could be possible to integrate [3] Y. Cai, G. Huang, et al., An optimization-model-based interactive decision supportfurther mechanics, e.g. stemming from games, to motivate users to system for regional energy management systems planning under uncertainty,participate in such a system. Expert Systems with Applications 36 (2) (2009) 3470–3482. [4] X. Kai, L. Yong-qi, Z. Zhi-zhong, Y. Er-keng, The vision of future smart grid, Electric Power 06 (2008).Acknowledgment [5] Lyn Bartram, Rodgers Johnny, Muise Kevin, Chasing the negawatt: visualization for sustainable living, IEEE Computer Graphics and Applications 30 (3) (May/June 2010) 8–14, doi:10.1109/MCG.2010.50. This research was supported by Basic Science Research Program [6] N. Motegi, M. Piette, et al., Web-based Energy nformation Systems for Energythrough the National Research Foundation of Korea (NRF) funded by Management and Demand Response in Commercial Buildings, 2003.the Ministry of Education, Science and Technology (2010-0023194). [7] E. Odum, Energy, ecosystem development and environmental risk, Journal of Risk and Insurance 43 (1) (1976) 1–16.This work is also partially supported by Korea Minister of Ministry of [8] J. Wong, J. Hong, What do we mashup when we make mashups? ACM (2008)Land, Transport and Maritime Affairs (MLTM) as U-City Master and 35–39.Doctor Course Grant Program. [9] [10] [11] [1] M. Amin, B. Wollenberg, Toward a smart grid: power delivery for the 21st century, IEEE Power & Energy Magazine 3 (2005) 34–41. Please cite this article as: S.A. Kim, et al., Integrated energy monitoring and visualization system for Smart Green City development, Autom. Constr. (2011), doi:10.1016/j.autcon.2011.07.004